I was about to post a "they're doing it wrong" comment, suggesting just this sort of thing. Interesting that the idea of applying vector network analysis to household power monitoring is so old that the patent has expired.

Using an optical sensor to monitor a spinning-wheel meter is just goofy, and so, for that matter, is the direct connection to the AC mains in the patent. All of the information they need can be sensed magnetically, by proximity to the house's power feed.

Your presuming that there is exposed access to the cable feeding the main panel, that is often not the case. I have no problem opening a mains panel or otherwise mucking around to get an inductive clamp on such a cable, but I don't think you can make the assumption about their target market. Their solution may seem 'clunky' from a technical standpoint, but it has one very important thing going for it: anyone can do it, without risk and without a knowledge of electricity, all they need is a conventional meter.

Counterpoint: the fact that they need an outlet to power the thing is bad, now the 'common Joe/Jane' doesn't need to mess with the electrical stuff but instead they need to drill a hole through the wall, near the meter, to get a wire to the sensor. I don't like the idea that the user to has a drill a hole near the mains entrance to their home, that can result in bad things.

You mean like this, http://salestores.com/tenma727218c.html?gclid=CLqlss7l5bECFW...

Something like that, yes. Combine that concept with the notion of the SWR bridge, and there's no reason you couldn't do vector analysis, rather than just scalar sensing.

Why not build what you describe, clamp it around the main feed to the house and have it join the wifi network, it can pubsub the data anywhere at that point.

This would be an order of magnitude better than just reading the meter.

Yeah, I think we'll see the power companies doing that themselves before too long, if they aren't already.